Abstract

The article is devoted to the analysis of the Shewhart control charts sensitivity for quantitative data depended on the number of sample values, on the basis of which the current values and are calculated, and depends on the probability of choosing the control limits. The sensitivity of the Shewhart control charts is their ability to detect possible changes in the technological process, i.e. the appearance of the process biasing or its values scattering increasing. Sensitivity analysis for mean values control charts was carried out. Sensitivity is estimated based on the mean length of the series – the mean number of samples. The samples are taken and analyzed through the same time intervals. It is supposed that the sample mean values are not correlated. As shown by the analysis, the sensitivity depends on the number of elements in the samples for mean value charts. In relation to corrective actions, the chart sensitivity of average values for the number of values in the sample is 20% is less than at . This can significantly increase the allowable losses due to the reduction of product quality produced in small batches. The dependence at is additionally presented to reflect the effect of the sample volume. S-charts sensitivity analysis is carried out. It indicates that assertion regarding the numerical values in the samples in the normative documents ( and in some cases 3) is fair. When using - and -charts together, the process is considered to be statistically manageable in the case when both charts confirm this. Practically the simultaneous shift of the normalized value of the technological process and scattering increase is almost not observed. Therefore, this case is more theoretical. The knowledge of the control charts sensitivity (the number of steps of sampling control) allows us to determine the intervals between samples and introduce additional warning boundary values, when it is necessary, provided that the technological process is short-lived (typical for low-power production). This is especially true for low-power enterprises, where the assortment of products varies relatively often and the correct construction of the control process is a guarantee of the products quality.

Authors and Affiliations